docs(gfql): q1-q9 board update to the candidate 20260803 lanes (17W/0T/1L) - #1848
Merged
Conversation
…7W/0T/1L)
Transcribe the two receipted candidate lanes (pyg-bench
results/graphbench-board-{20k,100k}-cand-20260803, PR #170) into the board:
all 72 graphbench cells move to the new runs, the 20k board reads 9W/0T/0L
(q8 a WEAK win: slot ranges overlap) and the 100k board 8W/0T/1L with q8 the
one loss (LOSE 1.52x), stated plainly. WEAK disclosures travel as data on
the two overlapping cells (q8@20k, q5@100k) and come off the cells that no
longer overlap (q4@20k, q5@20k). Provenance extends with the two new run
records pinned to the combined publication build 938f228 (master f875724
plus PRs #1844 #1845 #1846 #1847). Comparator tables transcribed verbatim;
no archaeology, facts only.
Verified: docs/test_bench_numbers.py 18 passed; rstcheck (repo config) clean
on the page.
Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_011AB4RZpph3uSFUpzKnZJcr
20k 9W/0T/0L (q8 WIN 1.48x clean; weak flag moves to q4) and 100k
7W/1T/1L (q8 LOSE 1.37x; q5 TIE inside the 10% band) on candidate build
5c1aee1b0, AST-identical to landed master. Cells, verbatim comparator
tables, prose, and provenance entries all regenerated from
pyg-bench results/graphbench-board-{20k-r3a-20260805,100k-r3a-20260806}.
Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01MF7uRZLKZaD6Q9FGWSmyXi
lmeyerov
marked this pull request as ready for review
August 6, 2026 04:04
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
1 participant
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
What
Follow-up to #1815: update the q1–q9 board section of
docs/source/gfql/performance.rstanddocs/source/_data/gfql_benchmarks.jsonto the two sealed candidate lanes at publication build938f22851e68918799da2bee49c88db2e2c103ee(masterf875724ce+ #1844 + #1845 + #1846 + #1847).DRAFT on purpose: this must not merge before the four perf PRs land — the page would otherwise quote numbers master cannot reproduce. Same page conventions as #1815: verbatim comparator transcription, WEAK disclosures carried as data, a single provenance block, topline stated plainly, no archaeology.
Changes
graphbench.*cells move to the new runsgraphbench-board-{20k,100k}-cand-20260803with the values from each lane'scompare.txt(pyg-bench PR graphistry/pyg-bench#170).LOSE 1.52x), stated plainly.graphbench.20k.q8.polars_vs_kuzuandgraphbench.100k.q5.polars_vs_kuzu(slot ranges overlap in the new lanes); removed fromgraphbench.20k.q4.polars_vs_kuzuandgraphbench.20k.q5.polars_vs_kuzu(no longer overlap).compare.txt.938f22851(candidate publication build: masterf875724ceplus PR perf(gfql): dense-domain proof kernel for the equal-domain two-hop count (q8) #1844, PR perf(gfql): drop domain semi-joins subsumed by property joins in the fused single-hop lane (q1/q3/q4) #1845, PR perf(gfql): minimal-join fused plan for the two-star grouped count (q5/q6/q7) #1846, PR perf(gfql): elide the interval shift in the dense-domain two-hop kernel (q8@100k, round 2) #1847; pyg-benchfae1e975, lanes published via graphistry/pyg-bench#170, perf lock held, VALIDITY clean in both lanes); the page's provenance block renders the two new runs. The 20260802 run records stay in the JSON as the master-pin baseline record.Verification
python3 -m pytest docs/test_bench_numbers.py— 18 passed (contract re-verification, staleness, reference/quotability, provenance/disclosure audits, chart drift — charts reference no graphbench cell, so no SVG regeneration is needed).rstcheck --config docs/.rstcheck.cfg docs/source/gfql/performance.rst— "Success! No issues detected."Do not merge before #1844, #1845, #1846, #1847. Owner reviews.
🤖 Generated with Claude Code
https://claude.ai/code/session_011AB4RZpph3uSFUpzKnZJcr